The research presented in this report is part of the regional habitat restoration program in the lower Columbia River and estuary (LCRE). As part of this program, we have established a suite of reference sites to help meet the goal of understanding and restoring wetland habitat. The data collected at these reference sites from 2005 through the present were analyzed in this study to meet two primary objectives: 1) to inform restoration planning and design by quantifying the ecological and hydrological conditions necessary for development of wetland plant communities and tidal channel networks and 2) to evaluate the effectiveness of wetland restoration actions in the LCRE by comparing restoration and reference site monitoring data. In this report, we present the results of the analysis of 51 reference wetland sites, focusing on the elevation, sediment, and inundation ranges required by native tidal wetland vegetation. We describe critical factors influencing existing wetland patterns in the LCRE, including the vegetation assemblages present, the elevation ranges at which they occur, and the inundation dynamics that result in their current distribution. Finally, we present how these data can be used to evaluate restoration action effectiveness. v Executive Summary vi Hydro-Vegetation Zones Shallow-water vegetation assemblages show distinct differences along the gradient between the mouth of the river and the upstream end of the estuary at Bonneville Lock and Dam. There are three zones based on species richness; the central region (rkm 50 to rkm 150) has the greatest number of species, and the upper and lower ends of the estuary have lower numbers of species. These three species richness zones can be characterized hydrodynamically as tidal-dominated, mixed tidal and riverdominated, and river-dominated, moving from the mouth of the Columbia River to Bonneville Dam. We hypothesize that fewer vegetation species are physiologically adapted to the extreme inundation in the upper end of the estuary, and, likewise, few are adapted to the tidal variability and salinity in the lower estuary. The fact that the mixed zone contains the greatest number of species suggests that the natural ecological disturbance regime may be lower there, and there may be a larger species pool adapted for these conditions in this zone. This intermediate disturbance hypothesis has been used in many ecosystems to describe the conditions that result in higher species diversity. Further examination of the hydrologic gradient revealed that the estuary can be divided further into five zones, driven primarily by salinity intrusion at the lower end and stronger fluvial flooding influence at the upper end. The breaks for these zones occur at approximately rkm 40, 104, 136, and 181. These breaks are preliminary and should be refined with additional data in areas of sparse sites and with other hydrologic analyses currently under way. The five hydro-vegetation zones developed from this analysis provide a means of determining the ranges of controlling factors ...

Abstract. This study adapts and applies the evidence-based approach for causal inference, a medical standard, to the restoration and sustainable management of large-scale aquatic ecosystems. Despite long-term investments in restoring aquatic ecosystems, it has proven difficult to adequately synthesize and evaluate program outcomes, and no standard method has been adopted. Complex linkages between restorative actions and ecosystem responses at a landscape scale make evaluations problematic and most programs focus on monitoring and analysis. Herein, we demonstrate a new transdisciplinary approach integrating techniques from evidence-based medicine, critical thinking, and cumulative effects assessment. Tiered hypotheses about the effects of landscape-scale restorative actions are identified using an ecosystem conceptual model. The systematic literature review, a health sciences standard since the 1960s, becomes just one of seven lines of evidence assessed collectively, using critical thinking strategies, causal criteria, and cumulative effects categories. As a demonstration, we analyzed data from 166 locations on the Columbia River and estuary representing 12 indicators of habitat and fish response to floodplain restoration actions intended to benefit culturally and economically important, threatened and endangered salmon. Synthesis of the lines of evidence demonstrated that hydrologic reconnection promoted macrodetritis export, prey availability, and juvenile fish access and feeding. Upon evaluation, the evidence was sufficient to infer cross-boundary, indirect, compounding, and delayed cumulative effects, and suggestive of nonlinear, landscape-scale, and spatial density effects. Therefore, on the basis of causal inferences regarding foodweb functions, we concluded that the restoration program is having a cumulative beneficial effect on juvenile salmon. The lines of evidence developed are transferable to other ecosystems: modeling of cumulative net ecosystem improvement, physical modeling of ecosystem controlling factors, meta-analysis of restoration action effectiveness, analysis of data on target species, research on critical ecological uncertainties, evidence-based review of the literature, and change analysis on the landscape setting. As with medicine, the science of ecological restoration needs scientific approaches to management decisions, particularly because the consequences affect species extinctions and the availability of ecosystem services. This evidencebased approach will enable restoration in complex coastal, riverine, and tidal-fluvial ecosystems like the lower Columbia River to be evaluated when data have accumulated without sufficient synthesis.

A lthough the common foundations of site-scale ecosystem restoration are well understood, the spatial scale and duration of restoration are rapidly expanding, raising theoretical questions and practical concerns. For instance, the primary goal of the Bonn Challenge, issued jointly in 2011 by the International Union for Conservation of Nature and the Government of Germany, is to restore 350 million ha of degraded land by 2030, while the UN General Assembly recently proclaimed 2021-2030 to be the Decade on Ecosystem Restoration. Such coordinated restoration across large spatial and temporal scales is a response to widespread environmental degradation, human welfare needs, and increased understanding of how species are sustained by distributed habitats and ecosystems (Lotze et al. 2006; Hall et al. 2018). In view of these trends, the Society for Ecological Restoration (SER) recently formed a Large-Scale Ecosystem Restoration section (Daoust et al. 2014). What does ecological restoration science offer to those working toward such ambitious goals? Restoration ecology provides information about the study of individual sites, ecosystems, and vulnerable species developed over the past half century (Roman and Burdick 2012; Clewell and Aronson 2013), yet for the most part it has not addressed large-scale restoration that includes multiple ecosystems and restoration projects across landscapes. Large-scale restoration is usually more cost-effective than local site-specific planning (Neeson et al. 2015); however, little formal research on achieving successful program-level outcomes has been reported. Useful principles to support the enormous projected expansion of restoration and ensure that large investments produce planned ecosystem functions are urgently needed. In practice, large-scale restoration is typically overseen by multidisciplinary teams and based on an ecosystem approach developed at the site scale, as can be seen in the programs we reviewed (Figure 1). Geomorphic conditions and hydrological

Sea otters (Enhydra lutris (L.)) were hunted to extinction off the coast of Washington State early in the 20th century. A new population was established by translocations from Alaska in 1969 and 1970, and currently numbers at least 550 animals. A major threat to the population is the ongoing risk of major oil spills in sea otter habitat. We apply population models to census and demographic data in order to evaluate the status of the population. We fit several density dependent models to test for density dependence and determine plausible values for the carrying capacity (K) by comparing model goodness of fit to an exponential model. Model fits were compared using Akaike Information Criterion (AIC). A significant negative relationship was found between the population growth rate and population size (r 2 ¼ 0.27, F ¼ 5.57, df ¼ 16, p < 0.05), suggesting density dependence in Washington state sea otters. Information Criterion statistics suggest that the Beverton-Holt model is the most parsimonious, followed closely by the logistic model. Values of K ranged from 612 to 759 with best-fit parameter estimates for the Beverton-Holt model including 0.26 for r and 612 for K. The latest (2001) population index count (555) puts the population at 87-92% of the estimated carrying capacity, above the suggested range for optimum sustainable population (OSP). Analyses of extinction risk associated with oil spills were based on assumptions of variable spill frequency, and variable mortality rates associated with spills once they occurred. At lower rates of assumed spill frequency, extinction risk was either negligible or was substantial only at the highest assumed mortality rates. At the highest assumed frequency rate (0.5 yr À1 ), extinction risks were high across the full range of assumed mortality rates. Elasticity analysis was conducted to examine the effects of proportional changes in vital rates on the population growth rate (l). The elasticity values indicate that the population is most sensitive to changes in survival rates (particularly adult survival).

Reservoir management on the Missouri River has changed the flow regime that once created dynamic emergent sandbar habitat (ESH) for the interior least tern (Sternula antillarum) and piping plover (Charadrius melodus). High flows that create large amounts of ESH are now rare, but the remaining interannual variability in river stage has strong effects on the amount of ESH available for nesting shorebirds. The scarcity of habitat has led the United States Army Corps of Engineers to develop an adaptive management plan for the restoration of ESH to support nesting terns and plovers. We describe the stochastic simulation models of ESH, plover populations and tern populations used in the adaptive management process, and examine the effects of river flow on projected outcomes of habitat restoration. The population models are most sensitive to uncertainty in adult survival rates. Model validation against historical amounts of ESH and population sizes suggests the model is a reasonable predictor of future dynamics. Flow variability contributes as much uncertainty as parameter estimation error to plover model projections but negligible uncertainty to the tern model. Autocorrelation in flow between years has stronger effects on population outcomes than the intensity of habitat restoration effort does. We compared population outcomes after a habitat-creating flow with population outcomes following habitat restoration and found that large pulses of habitat creation produced similar or better outcomes in the short term than low but consistent habitat restoration. However, bird populations fared better in the long term with low levels of restoration when habitat-forming flows were rare.